In order to improve service quality of DSN, it is necessary to consider coverage and energy consumption. On one hand, we should study how to improve area coverage quality by rotating sensing directions of sensor nodes, on the other hand, we should consider how to guarantee energy consumption balance by clustering network. This section reviews the existing area coverage optimization approaches and cluster-based energy consumption optimization approaches.
2.1. Area Coverage Optimization Approaches
Many optimization approaches have been proposed for area coverage problems. Among them, some explore the movement-assisted sensor node deployment by using mobile nodes to improve coverage quality [
7,
8,
9,
10,
11,
12], and these approaches are only suitable to the scenario where the sensor node is omnidirectional.
In Ref. [
7], the authors propose a virtual force algorithm as a sensor deployment strategy to enhance the coverage after an initial random placement of sensor nodes. In Ref. [
8], a localized scan-based movement-assisted sensor deployment method is proposed to use scan and dimension exchange to achieve a balanced state. To address another coverage problem, authors in Ref. [
9] propose the optimal coverage in directional sensor networks problem to cover maximal area while activating as few as sensor nodes as possible. To study how to place mobile sensor nodes to get high coverage, Ref. [
10] designs two sets of distributed approaches for controlling the movement of sensor nodes. In Ref. [
11], the authors propose a harmony search optimization algorithm to solve area coverage problem. A previous study in Ref. [
12] proposes a learning automata-based heuristic algorithm to find a near optimal solution to the proxy equivalent degree-constrained minimum-weight extension of the connected dominating set problem.
With the increasing application of surveillance camera and video sensing network, some approaches for directional sensor networks have been proposed. Some consider both the rotation and movement of sensor nodes [
13,
14,
15,
16,
17,
18], and the others only consider the rotation of static sensor nodes [
19,
20,
21,
22].
Ref. [
13] proposes a moving algorithm based on virtual forces of directional sensor nodes to eliminate coverage holes. In Ref. [
14], the authors present distributed self-deployment schemes of mobile sensor node according to the circumcenter and incenter of sensing direction. Taking another perspective on the coverage problem, authors in Ref. [
15] propose two Enahanced Deployment algorithms namly EDA-1 and EDA-2 to maximize coverage ratio for heterogeneous directional mobile network. To address the deployment problem, authors in Refs. [
16,
17] transform the area coverage problem into cell coverage problems by exploiting the Voronoi diagram. Ref. [
16] proposes a distributed approach to enhance the overall area coverage, and Ref. [
17] presents three local coverage optimization algorithm to improve coverage ratio. Different from above two approaches, Ref. [
18] presents several coverage increment algorithms namely vertex-based adjustment with Voronoi diagram (V-VD), edge-based adjustment with Voronoi diagram (E-VD), edge-based adjustment with Delaunay triangulation (E-DT) and angle-based adjustment with Delaunay triangulation (A-DT).
Authors in Ref. [
19] introduce the concept of sensing centroid into an omni-directional one, and the artificial fish-swarm algorithm is utilized to reduce the coverage hole and achieve the global coverage optimization. In Ref. [
20], the proposed approach studies how many sensor nodes are needed to meet a given required coverage probability under the circumstance that the sensing radius is adjustable. However, these two approaches ignore the redundancy ratio which is also a vital factor for improving coverage quality. In Refs. [
21,
22], the coverage ratio and redundancy ratio are considered at the same time. Among those, Ref. [
21] proposes a virtual potential field based coverage algorithm to increase the coverage ratio by forcing sensor nodes to turn form overlapping region to coverage holes after random deployment. In Ref. [
22], a virtual centripetal force-based coverage-enhancing algorithm is proposed to enhance the coverage by redeploying sensor nodes under the repel force-based centripetal force to shut off redundant nodes. However, these two approaches consider that coverage ratio and redundancy ratio have a negative correlation, and they are not suitable to the scenario where the target area is multiply covered.
Our proposed approach addresses the area coverage problem of DSN consisting of static and rotatable sensor nodes. With the objective of improving coverage ratio and reducing redundancy ratio by sensing direction rotation, we set up a multi-objective model in order to achieve high coverage ratio and low redundancy ratio. We next utilize an IAPSO to search for optimal sensing direction group to achieve multi-objective optimization for DSN. Compared to existing approaches [
21,
22], our proposed area coverage optimization approach can be suitable for multiply covered area where the negative correlation between coverage ratio and redundancy ratio is not obvious, and the IAPSO can effectively avoid local optima to some degree.
2.2. Cluster-Based Energy Consumption Optimization Approaches
A large number of cluster-based energy consumption optimization approaches have been proposed for energy consumption balance, and they can be categorized into two types: meta-heuristic approaches and nature-inspired approaches.
Meta-heuristic approaches are proposed on the basis of intuitive or empirical construction, and they usually have some randomness and uncertainty. The classic LEACH (Low-Energy Adaptive Clustering Hierarchy) is proposed in Ref. [
23]. LEACH can effectively reduce energy consumption to some extent, but it may select a cluster head with low residual energy and shorten the network lifetime. Therefore, many approaches have made some improvement on the basis of LEACH. HEED in Ref. [
24] selects cluster heads according to a hybrid of the node residual energy and a secondary parameter such as node proximity to its neighbors or node degree. PEGASIS in Ref. [
25] makes sensor node communicate only with a close neighbor and takes turns transmitting to the base station. In Ref. [
26], TL-LEACH saves energy consumption by using random rotation of local cluster base station. DL-LEACH in Ref. [
27] increases the energy efficiency of sensor node by reducing the transmission distance and simplifying the transmission routine for short-range transmission. In Ref. [
28,
29,
30], the authors propose energy efficient cluster based routing schemes for reliable networks. In Ref. [
31], E-LEACH uses homomorphic encryption to provide secure data aggregation and reduce energy consumption. According to Ref. [
32], M-LEACH is better than the TL-LEACH, and E-LEACH is better than M-LEAC. Although these approaches have better performance than LEACH, they cannot guarantee the energy consumption balance of sensor nodes when the size of the network is large.
Other approaches are proposed in the literature based on application of nature-inspired approaches. Among them, some are the improved LEACH. For example, LEACH-C in Ref. [
33] has better performance than LEACH because it considers the intra-cluster distance and residual energy of sensor nodes in the phase of cluster head selection. In Ref. [
34], a PSO based approach is proposed to select the optimal location of cluster head. Although these two approaches can guarantee the energy consumption balance of cluster head, they ignore the sink distance which is also an important factor to improve energy efficiency for direct communication of data to the base station. LEACH-FL takes battery level, distance and node density into consideration [
35], and it easily raises complexity and the accuracy problem in the fuzzification and defuzzification process.
As nature-inspired approaches which only utilize approximation algorithm to achieve energy consumption balance. PSO is utilized to achieve cluster head selection in Refs. [
36,
37,
38]. The authors in Ref. [
36] increase the network lifetime by reducing the total communication distance, and the authors in Ref. [
37] consider the intra-cluster distance and the residual energy of cluster head candidates. In Ref. [
38], the proposed approach considers the residual energy, intra-cluster distance, node degree and head count of the probable cluster heads. However, these three approaches cannot guarantee the performance with the size of the network verifying because they ignore the sink distance. In Ref. [
39], the authors propose an energy efficient clustering scheme based on recent variable population based chemo-inspired approach. To prolong the network lifetime, a PSO-based multiple-sink placement algorithm is proposed in Ref. [
40]. However, it ignores the fault-tolerance of a network.
Our proposed cluster-based approach belongs to a nature-inspired approaches. In the phase of cluster head selection, we consider the total residual energy and energy balance degree of the cluster head candidate nodes, and the two parameters can effectively guarantee the energy consumption balance of network whereas the existing approaches only consider the distance parameters or residual energy [
33,
34,
36,
37,
38]. We also propose a cluster formation algorithm in which the member nodes join in a cluster head by a weight function. However, in the existing approaches [
23,
33,
36,
37,
38], the member nodes join the cluster head by only considering distance, and it may cause imbalance load of cluster head energy consumption.
Based on the above analysis, area coverage optimization and energy consumption optimization are two key processes for improving service quality of DSN. However, they are usually studied in separate according to existing literatures. Most of the exiting approaches about area coverage optimization for DSN only focus on how to improve coverage ratio and ignore how to reducing redundancy ratio. Although some researchers consider the redundancy ratio, their proposed approaches are only suitable to the scenarios where the coverage ratio and redundancy ratio have the negative correlation. The existing cluster-based energy consumption optimization approaches cannot effectively guarantee energy consumption balance for network with different size because they usually only consider distance parameters and residual energy in the cluster head selection phase. In this paper, we take both area coverage quality and energy consumption into account and propose an approach based on IAPSO to achieve high coverage ratio and low redundancy ratio and energy consumption balance for DSN.
The contributions of this paper are as follows:
We propose a multi-objective area coverage optimization model which considers coverage ratio and redundancy ratio in order to reduce coverage blind areas and coverage redundant areas. This model is suitable for the scenario where the target area is multiply covered.
We propose a cluster head selection optimization model which considers the total residual energy ratio and energy balance degree of the cluster head candidate nodes to guarantee energy efficiency. We also propose an energy efficiency algorithm in the cluster formation phase.
We utilize an improved adaptive particle swarm optimization (IAPSO) to solve multi-objective area coverage optimization model and cluster head selection optimization model to achieve high coverage ratio, low redundancy ratio and energy consumption balance. Compared to traditional PSO, IAPSO has higher convergence ratio and operator precision.